Evaluation of the Quality of Evidence of the Association of Foods and Nutrients With Cardiovascular Disease and Diabetes: A Systematic Review

Abstract

Importance: Poor diet is a leading global factor associated with cardiometabolic disease (CMD). Understanding the quality of evidence of the associations between specific dietary factors and CMD, including effect size (relative risk [RR]) and uncertainty, is essential to guide policy and consumer actions to achieve healthy diet and public health goals.

Objective: To assess the quality of evidence of the associations between specific dietary factors and CMD as well as the quantitative evidence for RRs and the uncertainty of these risk estimates.

Evidence review: PubMed and the reference lists of eligible articles were searched between May 1, 2015, and February 26, 2021, for systematic reviews with meta-analyses of randomized clinical trials and prospective cohort studies that analyzed the consumption of 1 or more of the dietary factors of interest; reported dose-response meta-analyses; included healthy adults; and assessed 1 or more of the outcomes of interest. Study characteristics and RR estimates were extracted in duplicate. For identified associations, quality of evidence was assessed using the Bradford-Hill criteria for causation.

Findings: A total of 2058 potentially relevant reports were identified, from which 285 full-text articles were assessed for eligibility. The final selection of articles included 28 meta-analyses representing 62 associations between diet and CMD. Among these associations, 10 foods, 3 beverages, and 12 nutrients had at least probable evidence of associations with coronary heart disease, stroke, and/or diabetes. Most RRs ranged from 0.87 to 0.96 per daily serving change for protective associations and from 1.06 to 1.15 per daily serving change for harmful associations. Most identified associations were protective (n = 38) and a smaller number were harmful (n = 24), with a higher risk associated with higher intake.

Conclusions and relevance: This systematic review summarized the current quality of evidence of the associations of specific dietary factors with coronary heart disease, stroke, and diabetes. These findings may inform dietary guidance, the assessment of disease burden in specific populations, policy setting, and future research.

Figure 1.. Grading of Evidence of the Associations of Specific Dietary Factors With Cardiometabolic Outcomes

The 9 Bradford-Hill criteria for grading the evidence were strength, consistency, temporality, coherence, specificity, analogy, plausibility, biological gradient, and experiment. Each criterion is defined in eAppendix 1 in the Supplement. Gray indicates consistent evidence from well-designed studies with relatively few limitations; orange, consistent evidence from several well-designed studies with some important limitations; and brown, emerging evidence from a few studies or conflicting results from several studies. BMI indicates body mass index (calculated as weight in kilograms divided by height in meters squared); CHD, coronary heart disease; CVD, cardiovascular disease; MI, myocardial infarction; PUFA: polyunsaturated fatty acid; SBP, systolic blood pressure; and SFA, saturated fatty acid.

Figure 2.. Relative Risks (RRs) of the Associations of Foods With Risk of Cardiovascular Diseases (CVDs) and Type 2 Diabetes

Boxes in the plot show the RRs from the meta-analyses, and the horizontal lines through the boxes show the length of the 95% CIs. Associations supported by fewer than 3 studies that did not reference the individual studies were excluded. The number of RR estimates can be higher than the number of studies if there were more than 1 group in a randomized clinical trial, if estimates were separated by age or sex in prospective cohort studies, or if more than 1 prospective cohort study was included in a single study. Upper 95% CIs for association of fish or seafood with coronary heart disease (CHD) and stroke are significant (<1.0) when 3 significant digits are reported. MI indicates myocardial infarction; NR, not reported.

Figure 3.. Relative Risks (RRs) of the Associations of Beverages With Risk of Cardiovascular Diseases (CVDs) and Type 2 Diabetes and With Body Mass Index (BMI)

Boxes in the plot show the change in BMI (calculated as weight in kilograms divided by height in meters squared; A) and the RRs from the meta-analyses (B). The horizontal lines through the boxes show the length of the 95% CIs. Associations supported by fewer than 3 studies that did not reference the individual studies were excluded. The number of RR estimates can be higher than the number of studies if there were more than 1 group in a randomized clinical trial, if estimates were separated by age or sex in prospective cohort studies, or if more than 1 prospective cohort study was included in a single study. BMI indicates body mass index; CHD, coronary heart disease; NA, not applicable; NR, not reported; and SSB, sugar-sweetened beverage.

Figure 4.. Relative Risks (RRs) of the Associations of Nutrients With Risk of Cardiovascular Diseases (CVDs) and Type 2 Diabetes

Boxes in the plot show the RRs from the meta-analyses, and the horizontal lines through the boxes show the length of the 95% CIs. Associations supported by fewer than 3 studies that did not reference the individual studies were excluded. Number of RR estimates can be higher than the number of studies if there were more than 1 group in a randomized clinical trial, if estimates were separated by age or sex in prospective cohort studies, or if more than 1 prospective cohort study was included in a single study. CHD indicates coronary heart disease; NR, not reported; PUFA, polyunsaturated fatty acid; and SFA, saturated fatty acid.